Siesta 4.0

Release Notes for Siesta-4.0 (June 2016)

This version includes the van der Waals functionals, the new
load-balancing code for real-space grid operations, a Wannier90
interface, a new Orbital-Minimization-Method solver, and other
improvements and bug fixes that have been part of the development
version for some time. Its feature set has been frozen for several
months and it has undergone enough testing to be considered a
near-production version.

Please see the BACKWARD COMPATIBILITY ISSUES and the TECHNICAL NOTES
sections below for some important issues regarding backwards
compatibility with 3.X versions and previous development snapshots.

See the file REPORTING_BUGS for instructions to report bugs and
suggestions.

-*- NEW FEATURES

Please see the relevant section of the manual for more information.

* New SiestaXC library for exchange and correlation, implementing several
  van der Waals functionals and some newer GGA ones.

* New code to improve the load-balancing of the operations in the
  real-space grid when running in parallel.

* A new interface to the Wannier90 code for the generation of
  maximally localized wannier functions.

* New electronic-structure solver implementing the the Orbital-Minimization-Method

* New mixing options, including hamiltonian and charge-density mixing

* Charge-confinement and "filteret" basis-set generation options.

* Improved MPI version of the Siesta-as-subroutine code for executing
  independent calculations.

* New JobList utility to organize and run multiple jobs.

* Timer with call-tree awareness.

* Performance and usability enhancements in TranSiesta.

* Enhancements to the restart capabilities in molecular-dynamics runs.

* New options for wave-function output. New WFSX format.

* 'Fatbands' analysis.

* Enhancements to the COOP/COHP analyzer

* Hirshfeld and Voronoi charges. Bader analysis output

* Force-convergence diagnostics.

* New HSX file format for Hamiltonian/Overlap files

* Calculation of the vacuum level for non-bulk systems.

* Updates to other analysis tools in Util/

* Replacement of license-encumbered routines by new ones.

* Enhancements to the manual and build system.

In addition, a number of bugs have been fixed, and there have been
numerous cosmetic changes to the output and the code itself.

-------------------------
-*- BACKWARD COMPATIBILITY ISSUES

Please take into account the following changes in behavior (more
details in the TECHNICAL NOTES section below):

*** The grid functions (charge densities, potentials, etc) were in
single precision by default in the 3.X versions, but are in double
precision by default for post-3.X versions. The 'phi' array that holds
the values of the basis orbitals on the real-space grid is kept in
single precision. Please take this into account if you compare the
results with those of siesta-3.X runs. See the manual in both versions
for more information.

*** Changes in the geometry used for the analysis of the electronic
structure, as well as in the handling of the density matrix (DM) and
hamiltonian (H). This will _slightly_ change the output of most
calculations and the detailed results of any post-processing. Keep
this in mind if you need to maintain coherency within a project.

*** Electric field and dipole correction for slab calculations

Older versions applied an incorrect dipole correction when also using
an electric field (old behavior may be recovered by forcing
SlabDipoleCorrection to .false.)

Older versions over-estimated the energy contribution from the dipole
correction by a factor of 2 (old behavior cannot be recovered).

-------------------------
-*- TECHNICAL NOTES

*** Changes in the geometry used for the analysis of the electronic
structure, as well as in the handling of the density matrix (DM) and
hamiltonian (H).

-*-*-*-*-*
NOTE: This change will _slightly_ change the output
of most calculations and the detailed results of any
post-processing. Keep this in mind if you need to maintain coherency
within a project. See the rest of the text for ways to restore
backward compatibility.
-*-*-*-*-*

This change is part of a long-overdue clarification work related
to the "outer geometry loop" in Siesta and its interaction with the
analysis of the electronic structure.

The most important changes, following the actual refactoring logic, are:

* Superfluous calls to 'superx' to propagate the new coordinates to
  the auxiliary supercell in 'siesta_move' have been removed, as they
  are already issued by 'state_init'.

* The above change has the side effect of leaving inconsistent sets of
  unit cell and supercell coordinates in 'siesta_analysis' if the unit
  cell coordinates used in the latter are those 'moved' by the
  geometry-update routines. To avoid this, and to do the analysis
  using the actual coordinates used in the computation of the
  electronic structure, the coordinates are reset to those values
  (using the *_last arrays) in 'siesta_analysis'. The old behavior
  can be recovered by using by setting the backward compatibility
  switch 'compat-pre-v4-dynamics' to 'true'.

* To guarantee that the DM and H used in the analysis correspond to
  the latest electronic structure computed, there should be no mixing
  at the end of the scf loop. This behavior, which was optional until
  now (controlled by the value of scf-mix-after-convergence) is now
  the default. The use of DM_out instead of DM_mixed provides also a
  truly variational total-energy and more consistent forces. If the
  scf cycle is carried out mixing H, H(DM_in) is recovered by
  re-setting H to Hold in siesta_forces at the end of the cycle.

  Also, to preserve the original H for analysis it is necessary to
  avoid overwriting it in post_scf_work. This is now done by default,
  except if the option 'scf-recompute-h-after-scf' is 'true'.

  There is another switch for pre-v4.0 compatibility in this context:
  If 'compat-pre-v4-dm-h' is 'true', the 'scf-mix-after-convergence'
  and 'scf-recompute-h-after-scf' options are activated too.

  Some comments have been added to siesta_forces to document possible
  further clarifications.

* For clarity, each MD or geometry-optimization flavor has now its own
  self-contained logic block in 'siesta_move'. This has uncovered a
  few issues that have been corrected:

  - The quenched forms of the Verlet and Parrinello-Rahman (PR)
    schemes, which are actually relaxations, did not check for
    convergence before moving the atoms (PR did not check at any
    point). Now the 'constrained' forces and stress are checked
    before invoking the routines. In the case of PR, a 'target stress'
    can optionally be subtracted from the constrained stress (a
    further form of constraint that was previously only available
    (internally) in the standard relaxation routines). A future update
    might put all these constraint conditions in 'fixed', but care
    should be taken to study their interaction with the MD
    variable-cell routines. A future revision might look into the
    effectiveness of the variable-cell routines.
    There is a new example in Tests for quenched MD relaxation.

  - The geometry output in 'siesta_move' is now exclusively done for
    checkpointing purposes, and typically involves the 'moved'
    geometries. The current geometry output is done in 'state_init',
    and any final output in 'siesta_analysis'.

  - Some output calls have been removed where they did not make much
    sense (as in the FC or server blocks).

  - The deprecated 'phonon' support has been removed.

* Siesta_analysis outputs the 'current' final geometry, and not the
  'moved' one. The old behavior can be recovered by using the
  'compat-pre-v4-dynamics' switch.

* The default 'dynamics' option has been changed from 'verlet' to 'CG'.
  There should really be a new 'single-point' default which completely
  avoids 'siesta_move'. The old behavior can be recovered by using the
  'compat-pre-v4-dynamics' switch.

* Single-point calculations do not write .STRUCT_NEXT_ITER files, and
  the coordinates in the XV file are the current ones, unmoved.
  Extra output in siesta_options is avoided for this case.

* The headings for the beginning of the dynamics blocks are now
  more concrete.

*** Merge code for interoperation with Wannier90

(Javier Junquera, after work by Richard Korytar, and Alberto Garcia)

Siesta can now read the .nnkp file produced by Wannier90 and output
the *.amn, *.mmn, *.eig (see below), and UNK* files needed by
post-processing with that program, which generates maximally-localized
Wannier functions.

The most relevant changes in the code needed to support this
functionality are detailed below. For practical guidance, see the
manual, the simple test in Tests/wannier, and the suite of Wannier90
examples adapted for Siesta in

   http://personales.unican.es/junqueraj/Wannier-examples.tar.gz

* To implement support for "projection functions", which might not be
  pure radial functions, but hybrids with mixed angular momentum and
  varying spatial orientation, as well as to provide support of
  arbitrary kinds of radial functions for future developments, the
  'matel' framework has been extended:

  Matel (actually renamed to 'new_matel') now accepts two generalized
  indexes instead of the two pairs of indexes previously used. The
  generalized indexes can represent either a normal "radial function",
  as in previous Siesta usage, or a "projection function". The indexes
  are obtained via a "check-in" into a registry of functions, which
  keeps track of the nature of the functions and provides the needed
  evaluators and accessors for 'new_matel'.

  The relevant code is in 'matel_registry', 'new_matel', and
  'register_rfs'. The latter shows how to "register" the usual radial
  functions for PAOs, KB projectors, etc.

* Bloch states are computed and saved, instead of being recomputed as
  needed, to avoid problems with irreproducible random phases and
  linear combinations within degenerate subspaces.

* At this point the parallelization of the new code is only "over
  orbitals". It is not yet possible to parallelize the operations over
  the k-points on the grid. The production of the UNK* files can be
  slow if there are many k-points and bands.

* There is not yet support for "spinor wavefunctions". In the case of
  spin-polarized calculations, separate spin-up and spin-down
  processings are needed.

* There is not yet support for interaction with Wannier90 in "library
  mode", due to licensing issues.

* The "eigenvalue file" for Wannier90 has extension .eigW to avoid
  clashes with the standard Siesta eigenvalue file in case-insensitive
  filesystems.

*** Merge code for 'server' operation via sockets and i-PI interface

(Michele Ceriotti, Jose Soler)

* A new infrastructure for socket handling in f90 has been
implemented by Michele Ceriotti (EPFL), who is also the author
of the i-PI framework. From its web-page:

  i-PI is a Path Integral Molecular Dynamics (PIMD) wrapper code
  designed to be used together with ab initio evaluation of the
  interactions between the atoms. i-PI should be run as a separate
  program, and communicates with one or more instances of an
  electronic structure code using internet (or UNIX) sockets. The
  communication is regulated by a simple protocol, and requires
  patching the client code to handle the exchange of data with i-PI.

This revision implements the interface to the i-PI protocol. In
Siesta, the new socket functionality can also be used as a new form of
'server' operation.

See Util/Sockets for examples of the f90 socket interface,
Util/SiestaSubroutine for examples of the server operation (including
via sockets) and Tests/i-pi-test for a simple test of interoperability
with i-PI.

*** Fix shortcomings in reading of empirical potential block

  * The number of potential lines is now arbitrary.
  * Units handling has been improved, to allow arbitrary
    length and energy units.

  (Nick R. Papior)

*** Added restart file for smooth continuation of annealing runs

* Added the writing and reading of an .ANNEAL_RESTART file, along the
  same lines of the .VERLET_RESTART file. This fixes a small problem
  for restarting annealing calculations when a continuous smooth run
  is desired.

  (Fabiano Corsetti)

*** Updated gnubands (new.gnubands) and grid2cube (g2c_ng) programs
    in the Util section.

  (Nick R. Papior and Alberto Garcia)